Switching PSU advice

[John]

Hi !
I'm tryin' to service an IBM Server 325 PSU, suddenly dead.
I checked all capacitors, diodes and other stuff for shorts.
I put a paper clip connecting the power on pin with gnd.
The rectified voltage (generated by a separate board) reaches a primary
stage, with discrete components and three ICs
TOP200YAI - Three-terminal Off-line PWM Switch
LM393P - as usual
UC3843AN - Current Mode PWM Controller
CQY80NG - Optocoupler
and two transformers.

Well, from this stage exits nothing, with or w/o the paper clip.
My experience on switching PSUs is very limited.
The PowerOn signal doesn't reach directly the first stage but a custom made
riser card, so the fault could be elsewhere.
Any hints on where to begin lookin'?

As a last resort, since the output voltages are the same of a standard ATX,
only with different connectors, I could swap them and use a "normal" ATX.

But I'd give a try, it could be a useful experience...

 

[N_Cook]

Hi !
I'm tryin' to service an IBM Server 325 PSU, suddenly dead.
I checked all capacitors, diodes and other stuff for shorts.
I put a paper clip connecting the power on pin with gnd.
The rectified voltage (generated by a separate board) reaches a primary
stage, with discrete components and three ICs
TOP200YAI - Three-terminal Off-line PWM Switch
LM393P - as usual
UC3843AN - Current Mode PWM Controller
CQY80NG - Optocoupler
and two transformers.

Well, from this stage exits nothing, with or w/o the paper clip.
My experience on switching PSUs is very limited.
The PowerOn signal doesn't reach directly the first stage but a custom made
riser card, so the fault could be elsewhere.
Any hints on where to begin lookin'?

As a last resort, since the output voltages are the same of a standard ATX,
only with different connectors, I could swap them and use a "normal" ATX.

But I'd give a try, it could be a useful experience...

Where does tracing from the opto (control side) lead to?

 

[John]

A pic is better than words
http://imageshack.us/photo/my-images/211/img3469w.jpg/
http://imageshack.us/photo/my-images/255/img3468w.jpg/
http://imageshack.us/photo/my-images/253/img3467f.jpg/

 

[John]

I checked all primary ICs voltages: always 0.
PowerOn signals goes into a custom card, which seems to act as a controller.
nearby there's a current sense and other stuff.
Then, output signals from this custom card seem to go into the optocoupler,
to the primary side. No voltages around the optocoupler...
I begin to suspect the custom card.
But I can't desolder it. It's in a bad place, I need to desolder other stuff and
only with a small desolder pump ...

 

[N_Cook]

I checked all primary ICs voltages: always 0.
PowerOn signals goes into a custom card, which seems to act as a controller.
nearby there's a current sense and other stuff.
Then, output signals from this custom card seem to go into the optocoupler,
to the primary side. No voltages around the optocoupler...
I begin to suspect the custom card.
But I can't desolder it. It's in a bad place, I need to desolder other stuff and
only with a small desolder pump ...

So what hapens if you cut the control line to the opto and power the opto
LED on via a droppered battery for a moment.

 

[John]

So what hapens if you cut the control line to the opto and power the
opto LED on via a droppered battery for a moment.

You mean to supply +5V to the control input of the opto, having disconnected
that line from the custom card ?

 

[John]

There's a daughter card, who rectifies the main AC: the voltage across the two
big caps is correct.
Across source and drain of the TOP200YAI - Three-terminal Off-line PWM Switch
there is the rectified voltage. The control input is 0.

On the control line of the opto is not 0, but 0.34V.

It seems that the PSU lacks the "start signal".

 

[Jamie]

stuff and




So what hapens if you cut the control line to the opto and power the opto
LED on via a droppered battery for a moment.

I see comment about 0 volts all over it.. My question would be, is there
any AC coming in and rectified DC ? Switches, etc...

Kind of reminds of the IBM service desk trying to help some one with
their new computer. The screen is dark. After numerous steps given to
the customer to resolve issues, it was then ask if the customer could
look in the back to make sure everything was plugged in. When the
customer indicated he needed to get a flash light to do so, he then
commented about losing power to the house.

Of course, most here know the rest of that story, but it is a simple
matter of starting at the fuel pump and working your way to the tail
pipe.

Jamie

 

[Jamie]

There's a daughter card, who rectifies the main AC: the voltage across the two
big caps is correct.
Across source and drain of the TOP200YAI - Three-terminal Off-line PWM Switch
there is the rectified voltage. The control input is 0.

On the control line of the opto is not 0, but 0.34V.

It seems that the PSU lacks the "start signal".

You have a power good wire "PG" and a start wire "GREEN" which as to
be pulled to common.. ALso, depending the vintage, you may not have a
"GREEN" wire..

I don't remember the color code for the PG line, I think it was Orange
but don't quote me on that one. It'll be in the main bundle to the
PC board, not any of the hardware power connectors.

Jamie

 

[John]

Digital multimeter and ESR meter ...
I checked cabling against IBM manual.

 

[Kripton]

Digital multimeter and ESR meter ...
I checked cabling against IBM manual.

often with these 3842 controlers, I change the small cap (often 10uF)
around this chip between v+ and v-
it has very often solved the problem...
and it checks good on esr meter ...

but if you have no voltage around the chips it must not be that...

--

 

[John]

How about posting a link to that manual? '325' is a series of
servers, like 360 was a series of mainframes.

http://download.boulder.ibm.com/ibmdl/pub/systems/support/system_x/mt6016us.pdf

I followed wiring diagram and trasplanted connectors to a standard ATX
(285W vs 240W of the original). The faulty PSU has all the connector shown, but
MB lacks two of them, one for the 3.3V and for P12.
I found the corresponding wire on the ATX standard connector, except for the
+5V standby, present on the ATX, absent on the IBM PSU.
I tried the modified PSU and the server started.
But ... when I pressed again the power switch, it shuts off partially.
The fans spin in spurts, and leds flash, as the PowerOn signal (green wire
on the ATX, violet on the IBM PSU) doesn't reset completely.
I checked again my rewire and it's correct.
May be a different signal handling ... but it's not a real trouble, by now.
As soon as I power down, I disconnect it from the main AC, I don't need
a standby behaviour, by now ...

 

[[email protected]]

Hi !
I'm tryin' to service an IBM Server 325 PSU, suddenly dead.
I checked all capacitors, diodes and other stuff for shorts.
I put a paper clip connecting the power on pin with gnd.
The rectified voltage (generated by a separate board) reaches a primary
stage, with discrete components and three ICs
TOP200YAI - Three-terminal Off-line PWM Switch
LM393P - as usual
UC3843AN - Current Mode PWM Controller
CQY80NG - Optocoupler
and two transformers.

Well, from this stage exits nothing, with or w/o the paper clip.
My experience on switching PSUs is very limited.
The PowerOn signal doesn't reach directly the first stage but a custom made
riser card, so the fault could be elsewhere.
Any hints on where to begin lookin'?

As a last resort, since the output voltages are the same of a standard ATX,
only with different connectors, I could swap them and use a "normal" ATX.

But I'd give a try, it could be a useful experience...

Usually the AC input is filtered, then goes to a bridge rectifier.
The output of the bridge is filtered by one or more caps - typical
values start at 100uF, voltages often run up to 400 volts. If it
isn't obvious, that can pack quite a wallop. Remember, this is a live
ground - voltage is present on all points relative to earth ground.

The first step is to check the output voltage of the bridge - it
should be 1.4 x your nominal AC voltage. Since that is present, make
sure it is correct, then check the three ICs. Using the negative
terminal of the bridge for the negative lead of the dmm, check the
voltages on each pin of them. One of them MUST have a voltage to
start the whole process working.

With the components you list, odds are the TOP200YAI is the controller
for the standby supply, while the UC3843AN is the controller for the
main supply. But it's not impossible the roles are reversed.

A common design uses a high value resistor to drop the AC input
voltage down to the 20 volts or so the UC3843AN can tolerate on pin 7.

One suggestion - datasheets and application notes are available for
both. Study the schematics they contain and compare them to what you
see.

PlainBill

 

[legg]

Hi !
I'm tryin' to service an IBM Server 325 PSU, suddenly dead.
I checked all capacitors, diodes and other stuff for shorts.
I put a paper clip connecting the power on pin with gnd.
The rectified voltage (generated by a separate board) reaches a primary
stage, with discrete components and three ICs
TOP200YAI - Three-terminal Off-line PWM Switch
LM393P - as usual
UC3843AN - Current Mode PWM Controller
CQY80NG - Optocoupler
and two transformers.

Well, from this stage exits nothing, with or w/o the paper clip.
My experience on switching PSUs is very limited.
The PowerOn signal doesn't reach directly the first stage but a custom made
riser card, so the fault could be elsewhere.
Any hints on where to begin lookin'?

As a last resort, since the output voltages are the same of a standard ATX,
only with different connectors, I could swap them and use a "normal" ATX.

But I'd give a try, it could be a useful experience...

You'll need to get the small housekeeping power supply running, with
it's output voltage present, before the main one can be troubleshot.

The TOP200 integrated controller/switch will be responsible for this.
Look for rectifier/zener failures on output and around the small
transformer primary. It should be regulating the voltage on it's own
control pin 1.

RL

 

[Mike Tomlinson]

but if you have no voltage around the chips it must not be that...

Voltage to the 3842 is fed from the rectified mains via a high value
dropper resistor and clipped by a zener diode. I've had several cases
of that resistor going o/c or the zener going s/c.